A closer look at what is actually inside a transistor package illustrates some device considerations for 1 2 GHz device technology.

Figure 2 shows an open transistor, the Ericsson PTB 20105 (20 watts bipolar transistor, 925 960 MHz, 10 dB gain, 25 volts supply voltage), as seen by Scanning Electron Microscopy, SEM. The magnification is 27 times. This transistor is typically used in cellular base station amplifiers. The following parts can easily be identified:

1. The silicon chips In this particular transistor, two transistor chips (1.4 0.9 mm, 55 35 mils), containing four transistor cells each, are used in parallel (actually only 3+3 transistor cells are connected). The chips are mounted very precisely, close to the edge of the metallization area, to minimize critical bond wire lengths and to ensure the uniformity of the electrical characteristics. The silicon chips have been lapped to a thickness of around 120 m (<5 mils) from an initial thickness of 525 m (21 mils) to lower the thermal resistance to the package.

2. The chip capacitors for internal matching An input chip capacitor is connected to the input side (base), while no output matching was necessary for this device. A highly doped (n+) silicon is used for the chip capacitor, which is of MIM (metal-insulator-metal) type, with one metal being the highly doped substrate silicon and the other metal being a similar gold metallization as used on the transistor chips. The insulator is silicon dioxide or silicon dioxide + silicon nitride. The chip...